Establishing common ground when attempting to disambiguate spatial locations is difficult at the best of times, but is even more challenging between children and robots. Here, we present a study that examined how 94 children (aged 5-8) communicate spatial locations to other children, adults and robots in face-to-face interactions. While standard HRI implementations focus on non-ambiguous statements, we found this only comprised about 20% of children’s task based utterances. Rather, they rely on brief, iterative, repair statements to communicate about spatial locations. Our observations offer strong experimental evidence to inform future dialogue systems for robots interacting with children.